Video
Watch the video below. Just one image and an AI model are all you need to create a hyper-realistic video. There is no sound, but the visuals speak for themselves.
Source video: https://x.com/BenjaminDEKR/status/1829076521350312440
AI-generated videos are evolving rapidly. Here are just a few examples of what is already possible:
- AI video avatars: creating virtual characters with realistic gestures
- Face swapping (deepfakes): replacing one face in videos with another
- Image-to-video: turning still images into moving ones
- Text-to-video: generating videos from text descriptions
AI in videos is still in its early stages, which can sometimes lead to strange results. Yet, this technology is already capable of creating lifelike, expressive, and moving human figures.
Check out a few examples below. Not all videos have sound.
Since making videos with GenAI is expensive, offers a lot of variety, and is not yet widely accessible, you will not receive specific prompts in this chapter. Want to give it a try anyway? Feel free to use inspiration from the previous chapter.
Perhaps you will soon be creating your own videos to support your learning process, and they might even be useful for your teachers or lecturers. Imagine a video where a biology professor takes you on a virtual tour of a rainforest. You would discover how different plant and animal species interact, supported by animations and diagrams. This could significantly enhance your understanding of ecology. Who knows what else will be possible in the future.
For now, we still have to rely on slightly more handcrafted AI videos like this example, but did you know that GenAI is getting better and better at working with video? For instance, multimodal LLMs can process entire feature films through prompts, and just like with images, video-to-video models are emerging.